Right now we only support Anthropic models for summaries and only support a specific Groq model for narration improvement. It would be better if we supported a set of AI providers and supported selecting a model from any configured provider for any task.
Ideally we would find a library that does this for us and presents all of the AI providers via the same interface. Another option would be using the OpenAI API, since I think all providers support it.
One special case is if we need any provider-specific functionality, or functionality that can't be supported generically / through the OpenAI API. I don't think we currently do but we'd need to check.
Note: I don't really have any interest in other models so I probably won't implement this, but I'm documenting it to make it clear that I would accept a PR that does this.
Right now we only support Anthropic models for summaries and only support a specific Groq model for narration improvement. It would be better if we supported a set of AI providers and supported selecting a model from any configured provider for any task.
Ideally we would find a library that does this for us and presents all of the AI providers via the same interface. Another option would be using the OpenAI API, since I think all providers support it.
One special case is if we need any provider-specific functionality, or functionality that can't be supported generically / through the OpenAI API. I don't think we currently do but we'd need to check.
Note: I don't really have any interest in other models so I probably won't implement this, but I'm documenting it to make it clear that I would accept a PR that does this.